Problems & Solutions
Every problem, solution, and failed tactic linked as a case. The next time a similar problem appears, the agent knows what to try and what to skip.
Long-term Memory for AI
ShellBrain.
Agents forget. ShellBrain gives agents long-term memory. Grounded and retrieved exactly when it's needed. Works for Cursor, Claude, and Codex.
$
curl -L ShellBrain.ai/install | bash
Recall
One Command.
Inner Agents
Subconscious.
Working Agent
Visible
The agent you are using pulls one lever: shellbrain recall. It receives a compact brief and keeps coding; it never has to manage the machinery underneath.
Below the Surface
Knowledge Builder
Inner Agent
After sessions, it reads the evidence trail and turns useful work into durable memory: problems, solutions, failed tactics, facts, preferences, changes, and concepts.
Recall Agent
Inner Agent
At recall time, it searches, expands graph context, weighs what is current, and synthesizes the small brief the working agent actually needs.
Epistemology
A Graph Built for Recall.
Facts & Preferences
Durable knowledge about the repo. Facts are never silently overwritten—when something changes, the revision chain is immutable.
Evidence-Backed
Every memory traces back to the session transcript that produced it. If the evidence is ambiguous, the write is skipped.
Current Truth
Records are active, stale, superseded, or wrong. Retrieval respects that. Truth always stays current.
Ownership
Your brain. Your machine. Your knowledge.
ShellBrain runs locally. Nothing leaves your machine. No cloud. No outbound telemetry.
Every solution. Every failure. Every pattern. That's the most valuable thing you'll ever build. It belongs to you.